國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
𝛽𝛽3 represents the coefficient of independent variables, thus capturing the marginal effect on economic growth. 𝛼𝛼 and 𝜖𝜖 represent the intercept and error term respectively. 𝐶𝐶𝐿𝐿𝐺𝐺𝐶𝐶𝐺𝐺𝐿𝐿𝐶𝐶𝑇𝑇𝑥𝑥,𝑡𝑡 represents both Log of population, system of government, Cold War period and natural resources rent.
4.5 Statistical Results and Interpretation Fixed Effects
Since the author is measuring the impacts good governance and trade openness have on the level of economic development of twelve countries for a period of 45 years, the author employed the Fixed Effects estimation model. Throughout the years, social scientists have turned to linear fixed effects regression models as the primary method for causal inference with longitudinal data (Kim & Imai, 2016). The use of linear fixed effects regression models enables scholars to account for unobserved time-invariant confounders. In sum, fixed effects is used to measure the effect of variables that vary over time in order to deduce the
relationship between the explained and explanatory variables.
Table 15 (p. 63) presents the fixed effect estimation results of the hypothesis using six different models. In Models 1, 2 and 3, the author used country fixed effects while on Models 4, 5 and 6 the author employed both country and yearly fixed effects. Model 1 measured the estimated effect of good governance-𝛽𝛽1 and trade openness- 𝛽𝛽2 , both independent from each other, on the economies of Commonwealth Caribbean countries (Log GDP/Capita). The results showed that a unit increase in good governance and trade openness increases the GDP per capita of a country by 0.0029 and 0.0005 respectively. The constant 𝛼𝛼 and the
coefficients 𝛽𝛽1and 𝛽𝛽2 are all positive, indicating that good governance and trade openness, when measured independently, both have positive effect on the economies of countries.
Moreover, results also prove to be statistically significant at the 90% confidence level (𝛽𝛽1) and the 99% confidence level (𝛽𝛽2 and 𝛼𝛼).
To further test the effects of the explanatory variables, Model 2 expands on Model 1 by including the estimated interaction effect the interaction variable - Good Governance X Trade Openness have on the economies of Commonwealth Caribbean Countries. A slight change was observed. Contrary to Model 1, the coefficients 𝛽𝛽1and 𝛽𝛽2 are negative in Model 2, indicating that a unit increase in good governance and trade openness, negatively affects
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 63
the economies of countries by -0.0075 and -0.0007 respectively. However, looking at the coefficient 𝛽𝛽3 of the interaction variable, 𝛽𝛽3 (0.0001) is positive indicting that the interaction
DOI:10.6814/THE.NCCU.IMPIS.016.2018.A06
Note: *p<0.1, **p<0.05, ***p<0.01.
Standard Errors in parentheses.
Table 15. Fixed Effects Model: The effect good governance and trade openness has on the level of economic development of Commonwealth Caribbean Countries, 1972-2016. Dependent Variable: Log GDP/Capita
With Country Fixed Effects With Country and Yearly Fixed Effects
(1) (2) (3) (4) (5) (6)
Good Governance .0029*
(0.0016)
Trade Openness .0005***
(0.0001)
Log. Population -.0322
(.0211)
Natural Resources .0007
(.0008)
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 65
of good governance and trade openness has a positive effect on the economies of countries (Log GDP/Capita). Of all the variables, only the interaction variable 𝛽𝛽3 together with the constant 𝛼𝛼 were statistically significant at the 99% confidence level. In summary, Model 2 shows that even when good governance and trade openness are measured independently, their interaction is what positively boosts a country’s economy.
Model 3 builds even more on Model 2 by keeping all previous variables constant and adding population size, system of government, Cold War period and natural resource rents as controls. Having a larger population size, a parliamentary system of government and Cold War period, a country experiences less economic growth by 0.0322), 0.0040) and (-0.0221) respectively. As per rents from natural resources, the results showed that a unit increase in natural resources rent boosts a country’s economy by 0.007. The coefficient of the interaction variable 𝛽𝛽3 remained positive indicting that the interaction of good governance and trade openness has a positive effect on the economies of countries despite the controls.
Good governance, Cold War period and constant variables are statistically significant at the 99% confidence level. These results provide stronger support to the argument that even when controlling for population size, system of government, Cold War period and natural resource rents, a country experiences economic growth when they experience higher level of trade openness and good governance at the same time.
In order to account for the changes across the years, Models 4, 5 and 6 used country and yearly fixed effects. Despite accounting for country and yearly effects, the results in these Models resembles the results in the previous models. Results in Model 4 Showed that that a unit increase in good governance and trade openness increases the GDP per capita of a country by 0.0014 and 0.0005 respectively. The constant 𝛼𝛼 and the coefficients 𝛽𝛽1and 𝛽𝛽2 are all positive, indicating that good governance and trade openness, when measured
independently, both have positive effect on the economies of countries. Moreover, trade openness also prove to be statistically significant at the 99% confidence level.
Contrary to Model 4, the coefficients 𝛽𝛽1and 𝛽𝛽2 are negative in Model 5, indicating that a unit increase in good governance and trade openness, negatively affects the economies of countries by -0.0108 and -0.0008 respectively. However, looking at the coefficient 𝛽𝛽3 of the interaction variable 𝛽𝛽3 is positive (0.0001) indicting that the interaction of good
governance and trade openness has a positive effect on the economies of countries (Log GDP/Capita).Good governance is statistically significant at the 95% confidence interval while the interaction variable 𝛽𝛽 together with the constant 𝛼𝛼 were statistically significant at
DOI:10.6814/THE.NCCU.IMPIS.016.2018.A06
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the 99% confidence level. In summary, Model 5 shows that even when using country and yearly fixed effects, the interaction of good governance and trade openness positively boosts a country’s economy.
Model 6 employed country and yearly fixed effects on all explanatory and control variables. Having a larger population size, parliamentary system of government and Cold War period, a country experiences less economic growth by 0.1007), 0.0161) and (-0.0798) respectively. As per rents from natural resources, the results showed that a unit increase in natural resources rent boosts a country’s economy by 0.032. The coefficient of the interaction variable 𝛽𝛽3 remained positive indicting that the interaction of good governance and trade openness has a positive effect on the economies of countries despite the controls.
Good governance, population size, Cold War period, natural resources and constant variable are statistically significant at the 99% confidence level. These results provide stronger support to the argument that even when controlling for population size, system of
government, Cold War period and natural resource rents, a country experiences economic growth when there is higher level of trade openness and good governance occurring at the same time.
Ordinary Least Squares (OLS) with Panel-Corrected Standard Errors (PCSE) A major downfall of TSCS data is heteroscedasticity. Given the concerns of correlations across the TSCS data, the author employed the Ordinary Least Squares with Panel-Corrected Standard Errors model. OLS with PCSE take care of heteroscedasticity and provide accurate estimates of the variability of the OLS estimates (Beck & Katz, 1995).
Table 16 (p. 66) presents the results for models: 7 to 12. The results of the PCSE are consistent with the results of the Fixed Effects. In Models 7, 8 and 9, the author used country dummies while on Models 10, 11 and 12 the author combined both country and yearly dummies. Model 7 measured the estimated effect of good governance-𝛽𝛽1 and trade openness- 𝛽𝛽2 , both independent from each other, on the economies of Commonwealth Caribbean countries (Log GDP/Capita). The results showed that a unit increase in good governance and trade openness increases the GDP per capita of a country by 0.0025 and 0.0210 respectively.
The constant 𝛼𝛼 and the coefficients 𝛽𝛽1and 𝛽𝛽2 are all positive, indicating that good governance and trade openness, when measured independently, both have positive effect on the
economies of countries. Moreover, results also prove to be statistically significant at
‧
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 67
Table 16. OLS with PCSE Model: The Effect Good Governance and Trade Openness has on the level of economic development of Commonwealth Caribbean Countries, 1972-2016. Dependent Variable: Log GDP/Capita
with country dummies with country and year dummies
Good Governance .0025*
(.0014)
Trade Openness .0210**
(.0379 )
Trade Openness .0001* (.0000) 0.0001**
(0.0000)
Natural Resources .0004
(.0006)
Panel-Corrected Standard Errors in parentheses.
DOI:10.6814/THE.NCCU.IMPIS.016.2018.A06
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the 90% confidence level (𝛽𝛽1) and the 95% confidence level (𝛽𝛽2).
To further test the effects of the explanatory variables, Model 8 expands on Model 7 by including the estimated interaction effect the interaction variable - Good Governance X Trade Openness have on the economies of Commonwealth Caribbean Countries. A slight change was observed. Contrary to Model 7, the coefficients 𝛽𝛽1and 𝛽𝛽2 are negative in Model 8, indicating that a unit increase in good governance and trade openness, negatively affects the economies of countries by -0.0051 and -0.0001 respectively. However, looking at the coefficient 𝛽𝛽3 of the interaction variable, 𝛽𝛽3 (0.0001) is positive indicting that the interaction of good governance and trade openness has a positive effect on the economies of countries (Log GDP/Capita). Of all the variables, only the interaction variable 𝛽𝛽3 together with the constant 𝛼𝛼 were statistically significant at the 90% confidence level. In summary, Model 2 shows that even when good governance and trade openness are measured independently, their interaction is what positively boosts a country’s economy.
Model 9 builds even more on Model 8 by keeping all previous variables constant and adding population size, system of government, Cold War period and natural resource rents as controls. This model showed that having a larger population size, a country experiences less economic growth by (-0.0085). As per a parliamentary system of government, Cold War period and rents from natural resources, the results showed that a unit increase in these
factors boosts a country’s economy by 0.015, 0.0007 and 0.0004 respectively. The coefficient of the interaction variable 𝛽𝛽3 remained positive-0.0001, indicating that the interaction of good governance and trade openness has a positive effect on the economies of countries despite the controls. The interaction variable is statistically significant at the 95% confidence level while population size and constant variables are statistically significant at the 99% confidence level.
These results provide stronger support to the argument that even when controlling for
population size, system of government, Cold War period and natural resource rents, a country experiences economic growth when there is higher level of trade openness and good
governance occurring at the same time.
In order to account for the changes across the years, Models 10, 11 and 12 combined country and yearly effects. Despite accounting for country and yearly effects, the results in these Models resembles the results in the previous models. Results in Model 10 Showed that that a unit increase in good governance and trade openness increases the GDP per capita of a country by 0.0030 and 0.0002 respectively. The coefficients 𝛽𝛽1and 𝛽𝛽2 are all positive, indicating that good governance and trade openness, when measured independently, both
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 69
have positive effect on the economies of countries. Moreover, both variables also prove to be statistically significant at the 95% confidence level.
Contrary to Model 10, the coefficients 𝛽𝛽1and 𝛽𝛽2 are negative in Model 11, indicating that a unit increase in good governance and trade openness, negatively affects the economies of countries by -0.0070 and -0.0008 respectively. However, looking at the coefficient 𝛽𝛽3 of the interaction variable 𝛽𝛽3 is positive (0.0001) indicting that the interaction of good
governance and trade openness has a positive effect on the economies of countries (Log GDP/Capita). The interaction variable 𝛽𝛽3 together with the constant 𝛼𝛼 are statistically significant at the 99% confidence level. In summary, Model 11 shows that even when using country and yearly fixed effects, the interaction of good governance and trade openness positively boosts a country’s economy.
Model 12 employed country and yearly fixed effects on all explanatory and control variables. Having a larger population size, a parliamentary system of government, and during Cold War period, a country experiences less economic growth by (-0.1142), (-.0190) and (-.0758) respectively. As per a and rents from natural resources, the results showed that a unit increase in these factors boosts a country’s economy by 0.0033 respectively. The coefficient of the interaction variable 𝛽𝛽3 remained positive (2.9085), indicating that the interaction of good governance and trade openness has a positive effect on the economies of countries despite the controls. All the coefficients, with the exception of good governance and
parliamentary system of government, are statistically significant at the 99% confidence Good governance statistically significant at the 95% confidence level. These results provide
stronger support to the argument that even when controlling for population size, system of government, Cold War period and natural resource rents, a country experiences economic growth when there is higher level of trade openness and good governance occurring at the same time.
To further facilitate the comprehension and interpretation of the research hypothesis and that of the interaction model, the author reproduced three graphical figures: Figure 5 (p.
69), Figure 6 (p. 69) and Figure 7 (p. 70). In Figure 5, we can observe that a country’s economic growth (GDP/Capita) increases when both, trade openness and good governance increase concurrently. However, countries with low trade openness (example: 50%) together with low good governance have a lower probability of experiencing high economic growth in the near future. Conversely, countries with high trade openness (example: 300%) together with high good governance have a higher probability of experiencing high economic growth
DOI:10.6814/THE.NCCU.IMPIS.016.2018.A06
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
in the near future. This graph further supports the author’s argument that countries with high trade openness and high good governance experience higher economic growth when
compared to countries with low trade openness and low good governance.
Figure 5. Predictive margins of trade openness and good governance on GDP/Capita Figure 6 and Figure 7 (p. 70) illustrate the different marginal effects the variables have on GDP/Capita. In Figure 6, the solid sloping line illustrates how the marginal effect of Trade openness changes with the score of good governance. The 95% confidence interval
Figure 6. The marginal effect of trade openness on GDP/Capita
56789Linear Prediction
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Good Governance
Trade Openness = 50% Trade Openness = 100%
Trade Openness = 150% Trade Openness = 200%
Trade Openness = 250% Trade Openness = 300%
Mean of Good Governance
0.1.2.3 Kernel Density Estimate of Good Governance
-.004-.0020.002.004Marginal Effect of Trade Openness
0 5 10 15
Good Governance
Thick dashed lines give 95% confidence interval.
Thin dashed line is a kernel density estimate of Good Governance.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 71
around the solid sloping line allow us to determine the conditions under which trade openness has a statistically significant effect (both above and below the zero line) on GDP/Capita.
Trade openness has a small negative effect on GDP/Capita whenever a country has very low good governance score. As predicted, this scenario changes to a positive effect whenever good governance score increases. Although there are countries with low good governance score, the mean score of good governance being 12 indicates that most countries in the study have high good governance score.
In Figure 7 we can observe the marginal effect good governance has on GDP/Capita.
The solid sloping line illustrates how the marginal effect of good governance changes with the level of trade openness. The 95% confidence interval around the solid sloping line allow us to determine the conditions under which good governance has a statistically significant effect on GDP/Capita. Good governance has a small positive effect on GDP/Capita whenever a country has very low trade openness, but this changes to a higher positive effect whenever the level of trade openness increases. From the graph, we can also observe that the level of trade openness has a normal distribution with a mean of 110%.
Figure 7. The marginal effect of good governance on GDP/Capita
Mean of Trade Openness
0.005.01.015 Kernel Density Estimate of Trade Openness
0.05.1.15Marginal Effect of Good Governance
50 100 150 200 250
Trade Openness (%)
Thick dashed lines give 95% confidence interval.
Thin dashed line is a kernel density estimate of Trade Openness.
DOI:10.6814/THE.NCCU.IMPIS.016.2018.A06
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Chapter Five: Conclusion
5.1 Summary of Research
Caribbean leaders are cognizant of the relatively small sized economies they possess compared to their neighbouring counterparts, and as such have sought for economic and political regional integration and globalization at large. Integration would ensure cooperation between member states and thus prepare them for a competitive world. To tap into the opportunity, Commonwealth Caribbean Countries have grown and diversified their
economies to survive in the competing world markets. Even after various trade agreements in place, similar economies, and so many similar characteristics, there exists varying differences in the economic growth trend amongst Commonwealth Caribbean countries since
post-independence. So, why did these twelve countries all experienced different economic growth patterns? What explains the different economic growth trends amongst the twelve
Commonwealth Caribbean Countries?
Given the Caribbean Commonwealth country’s close similarities, in various aspects and region’s high rate of government corruption (Transparency International, 2017), the researcher posits that trade openness alongside good governance play a key role in fostering sustained economic growth and development and consequently leading to an positive shift in the living standards of the twelve Commonwealth Caribbean countries. In trying to answer the research question, together with an under-researched literature, the author argues that the Commonwealth Caribbean countries experience different economic growth because of the different level of trade openness and good governance. The researcher’s hypothesis is that the degree of growth and development of the Commonwealth Caribbean countries will increase with the increase of both trade openness and good governance.
The hypothesis is examined through Time-Series Cross-Sectional (TSCS) analysis (panel data analysis) of trade openness and good governance on the economies of the 12 Commonwealth Caribbean countries for the period 1972 to 2016, (44 years). The period 1972-2016 is very important as it covers the post-independence period of most
Commonwealth Caribbean countries.
The author used fixed effects model and PCSE OLS to estimate the effects.. These results are consistent despite the model. Both good governance and trade openness influence the economic growth of countries.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TRADE OPENNESS, GOOD GOVERNANCE AND ECONOMIC GROWTH 73